iRadiology最新文献

{"title":"Ultrasmall Superparamagnetic Iron Oxide Nanoparticles as Positive Contrast Agents in Low-Field Magnetic Resonance Imaging","authors":"Ogechukwu Ngwu-Hyacinth,&nbsp;Aaron Alford,&nbsp;Caleb S. Rowe,&nbsp;Ryan Willoughby,&nbsp;S. Abdollah Mirbozorgi,&nbsp;Steven A. Rothenberg,&nbsp;Jesse Jones,&nbsp;Abin Sajan,&nbsp;Venkatesh P. Krishnasamy,&nbsp;Mark Bolding","doi":"10.1002/ird3.70035","DOIUrl":"https://doi.org/10.1002/ird3.70035","url":null,"abstract":"<p>Ultrasmall superparamagnetic iron oxide nanoparticles (usSPIONs) are promising alternatives to gadolinium-based contrast agents for positive contrast enhancement in magnetic resonance imaging (MRI). Unlike larger SPIONs that primarily function as T2/T2* negative contrast agents, usSPIONs with core diameters below 5 nm can effectively shorten T1 relaxation times, producing bright signals in T1-weighted images. This distinct behavior stems from their unique magnetic properties, including single-domain configurations, surface spin canting, and rapid Néel relaxation dynamics, which are particularly enhanced at low magnetic field strengths. The biocompatibility of iron oxide, efficient renal clearance pathways, and versatility for surface functionalization offer potential advantages over gadolinium-based agents, especially regarding safety concerns related to nephrogenic systemic fibrosis and gadolinium deposition. These nanoparticles show particular promise for applications in low-field MRI, vascular imaging, targeted molecular imaging, and theranostic platforms. Although challenges remain in optimizing synthesis methods for consistent production of monodisperse usSPIONs with tailored surface chemistry, ongoing research continues to advance their potential for clinical translation. This review explores the mechanisms, synthesis approaches, applications, and future perspectives of usSPIONs as positive contrast agents in MRI.</p>","PeriodicalId":73508,"journal":{"name":"iRadiology","volume":"3 5","pages":"372-387"},"PeriodicalIF":0.0,"publicationDate":"2025-09-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1002/ird3.70035","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145385181","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"The Value of [18F]FDG PET/CT for Differentiating Neoplastic From Benign Thrombus","authors":"Xiaofeng Dou,&nbsp;Congcong Yu,&nbsp;Chenxi Xue,&nbsp;Jing Wang,&nbsp;Chentao Jin,&nbsp;Yan Zhong,&nbsp;Rui Zhou,&nbsp;Xiaohui Zhang,&nbsp;Hong Zhang","doi":"10.1002/ird3.70036","DOIUrl":"https://doi.org/10.1002/ird3.70036","url":null,"abstract":"<div>\u0000 \u0000 \u0000 <section>\u0000 \u0000 <h3> Background</h3>\u0000 \u0000 <p>Vascular thrombus is a common complication in cancer patients and can be classified as benign or neoplastic based on tumor cell presence. Conventional imaging modalities provide anatomical information but exhibit limited specificity in distinguishing neoplastic from benign thrombi, prompting the evaluation of 2-deoxy-2-[¹⁸F]fluoro-D-glucose positron emission tomography/computed tomography ([¹⁸F]FDG PET/CT), which leverages metabolic differences for improved differentiation. In this study, the efficacy of [¹⁸F]FDG PET/CT for differentiating neoplastic thrombus from benign thrombus was evaluated in patients with malignant tumors.</p>\u0000 </section>\u0000 \u0000 <section>\u0000 \u0000 <h3> Methods</h3>\u0000 \u0000 <p>Sixty-five patients with histologically or clinically confirmed neoplastic or benign thrombus who underwent [¹⁸F]FDG PET/CT were retrospectively enrolled. Both visual and semi-quantitative analyses of the PET/CT scans were conducted. The difference in maximum standardized uptake value (SUV_max) between neoplastic and benign thrombus was assessed using a <i>t</i>-test. Furthermore, receiver operating characteristics (ROC) analysis was performed to determine the optimal SUV_max threshold for differentiating neoplastic from benign thrombus.</p>\u0000 </section>\u0000 \u0000 <section>\u0000 \u0000 <h3> Results</h3>\u0000 \u0000 <p>Within the cohort, 55 patients were diagnosed with neoplastic thrombus and 10 patients with benign thrombus. In the visual analysis, the sensitivity, specificity, positive predictive value, negative predictive value, and diagnostic accuracy of [¹⁸F]FDG PET/CT for diagnosing neoplastic thrombus were 96.4%, 90.0%, 98.1%, 81.8%, and 95.4%, respectively. In the semi-quantitative analysis, the sensitivity, specificity, positive predictive value, negative predictive value, and diagnostic accuracy for diagnosing neoplastic thrombus were 96.4%, 100%, 100%, 83.3%, and 96.9%, respectively. The normalized SUV_max of neoplastic thrombus was significantly higher than that of benign thrombus (10.11 ± 5.32 vs. 2.21 ± 0.51, <i>p</i> &lt; 0.001). The area under the ROC curves for visual assessment and semi-quantitative analysis were 0.932 and 0.993, respectively. There was no statistically significant difference observed between these two assessment methods (<i>p</i> = 0.317).</p>\u0000 </section>\u0000 \u0000 <section>\u0000 \u0000 <h3> Conclusions</h3>\u0000 \u0000 <p>[¹⁸F]FDG PET/CT is capable of differentiating neoplastic thrombus from benign thrombus. Both visual and semi-quantitative analyses demonstrated high diagnostic s","PeriodicalId":73508,"journal":{"name":"iRadiology","volume":"3 5","pages":"353-361"},"PeriodicalIF":0.0,"publicationDate":"2025-09-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1002/ird3.70036","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145385180","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Primary Duodenal Squamous Cell Carcinoma on 18F-Flurodeoxyglucose Positron Emission Tomography/Computed Tomography","authors":"Yinting Hu,&nbsp;Lei Jiang","doi":"10.1002/ird3.70032","DOIUrl":"https://doi.org/10.1002/ird3.70032","url":null,"abstract":"<p>A 48-year-old woman complained with repeated vomiting and pallor over 3 months. Laboratory results revealed a decrease in hemoglobin (80 g/L, reference range &gt; 110 g/L), and no other findings (including tumor markers) were abnormal. Next, abdomen computed tomography (CT) revealed a duodenal mass, which may be malignant. To further define the nature and stage of the lesion, the patient underwent ¹⁸F-flurodeoxyglucose positron emission tomography/computed tomography (¹⁸F-FDG PET/CT) scan. The maximum intensity projection (MIP) image (Figure 1a) demonstrated a high radioactivity (arrow) in the middle abdomen. The axial (Figure 1b), coronal (Figure 1c), and sagittal (Figure 1d) images of the abdomen displayed a solid lesion (arrow) with the size of 61 mm × 34 mm and a SUV_max of 15.6 in the horizontal part of the duodenum. In addition, peripheral lymph nodes with the maximum size of 8 mm × 5 mm showing mild FDG activity (SUV_max: 2.0) were noted (images not shown). The patient received surgical resection of the duodenal lesion (Figure 2a) and peripheral lymph nodes. Pathological examination (Figure 2b, hematoxylin–eosin staining and original magnification: ×100) from the duodenal specimen showed tumor cells arranged in nests with formation of keratin pearls. Immunohistochemistry indicated positive staining for P40 (Figure 2c, original magnification: ×100). These findings were consistent with a diagnosis of duodenal squamous cell carcinoma (SCC). Besides, peripheral lymph node metastases were also confirmed.</p><p>SCC of the duodenum is extremely rare and is more likely to represent metastasis from primary SCC originating in other sites, such as the head and neck, lungs, or cervix. Only occasional cases of primary SCC of the duodenum have been reported in the literature. There are four possible pathogenesis of primary duodenal SCC: (1) malignant transformation of heterotopic squamous epithelium; (2) pluripotential stem cells differentiate to malignant squamous cells; (3) squamous metaplasia malignant change due to chronic mucosal damage; and (4) adenocarcinoma transformed into adenosquamous carcinoma and eventually to SCC. Surgery might be the cornerstone in the management of such kind disease. Given the value of the differentiation of metastatic or primary duodenal SCC for treatment options and prognosis, establishing a correct diagnosis is essential. Traditionally, CT scanning has been the major imaging modality for diagnosing abdominal malignancies, which is not particularly sensitive for detecting duodenal malignancies. Increased FDG uptake in the duodenum is not uncommon, but it is usually physiological or inflammation-related. However, our case suggests that duodenal malignancy should be considered in the differential diagnosis when focal abnormal FDG uptake is present in the duodenum, especially when accompanied by a corresponding mass-like lesion on CT imaging. Furthermore, in this case, whole bo","PeriodicalId":73508,"journal":{"name":"iRadiology","volume":"3 4","pages":"315-317"},"PeriodicalIF":0.0,"publicationDate":"2025-08-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1002/ird3.70032","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144910069","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Evaluating the Capability of Large Language Model Chatbots for Generating Plain Language Summaries in Radiology","authors":"Pradosh Kumar Sarangi,&nbsp;Pratisruti Hui,&nbsp;Himel Mondal,&nbsp;Debasish Swapnesh Kumar Nayak,&nbsp;M. Sarthak Swarup,&nbsp; Ishan,&nbsp;Swaha Panda","doi":"10.1002/ird3.70030","DOIUrl":"https://doi.org/10.1002/ird3.70030","url":null,"abstract":"<div>\u0000 \u0000 \u0000 <section>\u0000 \u0000 <h3> Background</h3>\u0000 \u0000 <p>Plain language summary (PLS) are essential for making scientific research accessible to a broader audience. With the increasing capabilities of large language models (LLMs), there is the potential to automate the generation of PLS from complex scientific abstracts. This study assessed the performance of six LLM chatbots: ChatGPT, Claude, Copilot, Gemini, Meta AI, and Perplexity, in generating PLS from radiology research abstracts.</p>\u0000 </section>\u0000 \u0000 <section>\u0000 \u0000 <h3> Methods</h3>\u0000 \u0000 <p>A total of 100 radiology abstracts were collected from PubMed. Six LLM chatbots were tasked with generating PLS for each abstract. Two expert radiologists independently evaluated the generated summaries for accuracy and readability, with their average scores being used for comparisons. Additionally, the Flesch–Kincaid (FK) grade level and Flesch reading ease score were applied to objectively assess readability.</p>\u0000 </section>\u0000 \u0000 <section>\u0000 \u0000 <h3> Results</h3>\u0000 \u0000 <p>Comparisons of LLM-generated PLS revealed variations in both accuracy and readability across the models. Accuracy was highest for ChatGPT (4.94 ± 0.18) followed by Claude (4.75 ± 0.31). Readability was highest for ChatGPT (4.83 ± 0.27) followed by Perplexity (4.82 ± 0.29). The Flesch reading ease score was highest for Claude (62.53 ± 10.98) and lowest for ChatGPT (40.10 ± 11.24).</p>\u0000 </section>\u0000 \u0000 <section>\u0000 \u0000 <h3> Conclusion</h3>\u0000 \u0000 <p>LLM chatbots show promise in the generation of PLS, but performance varies significantly between models in terms of both accuracy and readability. This study highlights the potential of LLMs to aid in science communication but underscores the need for careful model selection and human oversight.</p>\u0000 </section>\u0000 </div>","PeriodicalId":73508,"journal":{"name":"iRadiology","volume":"3 4","pages":"289-294"},"PeriodicalIF":0.0,"publicationDate":"2025-08-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1002/ird3.70030","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144910059","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Applications and Implications of ChatGPT and GPT-4 in Radiology","authors":"Bo Gao,&nbsp;Weihua Ou","doi":"10.1002/ird3.70031","DOIUrl":"https://doi.org/10.1002/ird3.70031","url":null,"abstract":"&lt;p&gt;Rapid advancements in artificial intelligence (AI) technology have resulted in the emergence of state-of-the-art large language models (LLMs) such as ChatGPT and GPT-4. Originally designed for natural language processing, these models are now being applied to increasingly broader domains, particularly in medical image processing [&lt;span&gt;1&lt;/span&gt;]. Concurrently, the rise of such models has introduced innovative tools into medical image processing and diagnosis, profoundly shaping the future trajectory of this field. These tools not only enhance diagnostic accuracy and efficiency, but also alleviate substantial repetitive workloads for clinicians [&lt;span&gt;2&lt;/span&gt;]. To address the critical needs for transparency, reproducibility, and clinical reliability in biomedical AI research, Gallifant et al. [&lt;span&gt;3&lt;/span&gt;] proposed Transparent Reporting of a prediction model for Individual Prognosis or Diagnosis-LLM, an extension to the Transparent Reporting of a prediction model for Individual Prognosis or Diagnosis + artificial intelligence statement. In a domain-specific innovation, Liu et al. [&lt;span&gt;4&lt;/span&gt;] developed Radiology-GPT through training and fine-tuning on a massive radiology knowledge corpus. In comparison with general-purpose LLMs, this specialized model demonstrated superior performance, validating the feasibility of creating localized generative models for specific medical specialties. Complementing this work, Yuan et al. [&lt;span&gt;5&lt;/span&gt;] systematically evaluated the capabilities of the advanced multimodal model ChatGPT-4V for diagnosing brain tumors on 7T magnetic resonance imaging (MRI) data. Their study established a benchmark framework for ultra-high field imaging AI applications, propelling the progress of precision medicine and intelligent diagnostics.&lt;/p&gt;&lt;p&gt;This special issue on ChatGPT and GPT-4 includes four recent studies that cover applications of different LLMs, such as Meta LLaMA 3.1, ChatGPT, Claude, Gemini, and LLaVA, in various medical scenarios. Yuan et al. [&lt;span&gt;6&lt;/span&gt;] deeply explored the application of the Transformer architecture in natural language processing of chest X-ray reports, finding that this architecture holds significant potential in medical text processing. However, computational efficiency and ethical compliance require optimization, and future integration with multimodal data is needed to enhance diagnostic accuracy. Lotfian et al. [&lt;span&gt;7&lt;/span&gt;] evaluated the performance of the open-source model LLaMA 3.1 in thoracic imaging diagnostics using 126 multiple-choice questions. The model achieved an overall accuracy of 61.1%, with excellent performance in intensive care (90%) and terminology recognition (83.3%) but weaker results in basic imaging (40%) and lung cancer diagnosis (33.3%). This assessment demonstrates the potential of open-source models like LLaMA 3.1 while highlighting the need for domain-specific fine-tuning to improve stability as well as the need to balance open-source flexibility wit","PeriodicalId":73508,"journal":{"name":"iRadiology","volume":"3 4","pages":"259-260"},"PeriodicalIF":0.0,"publicationDate":"2025-07-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1002/ird3.70031","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144910324","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Four-Dimensional Analysis of F-18 Fluorodeoxyglucose Distribution","authors":"Nagara Tamaki,&nbsp;Tomoya Kotani,&nbsp;Yoshitomo Nakai,&nbsp;Kei Yamada","doi":"10.1002/ird3.70014","DOIUrl":"https://doi.org/10.1002/ird3.70014","url":null,"abstract":"<p>Positron emission tomography (PET) with F-18 fluorodeoxyglucose (FDG) is an established tool in clinical molecular imaging for evaluating metabolic activity. Advanced PET systems enable high-resolution three-dimensional analysis of FDG distribution, offering insightful information for lesion characterization. Furthermore, dynamic whole-body imaging facilitates assessment of temporal FDG uptake following its administration. This development introduces the concept of four-dimensional (4D) FDG PET/CT, enhancing the precision of lesion characterization. A critical advantage of applying 4D FDG PET/CT is the ability to distinguish pathological from physiological FDG uptake by analyzing motion-related uptake. Correcting motion and integrating serial imaging data enable accurate interpretation. Additionally, FDG uptake quantification is possible using Patlak analysis. This review discusses innovative clinical applications and quantitative techniques for analyzing FDG distribution through 4D FDG PET/CT.</p>","PeriodicalId":73508,"journal":{"name":"iRadiology","volume":"3 5","pages":"330-336"},"PeriodicalIF":0.0,"publicationDate":"2025-07-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1002/ird3.70014","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145385090","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"AI-Enhanced Predictive Imaging in Precision Medicine: Advancing Diagnostic Accuracy and Personalized Treatment","authors":"Aswini Rajendran,&nbsp;Rithi Angelin Rajan,&nbsp;Saranya Balasubramaniyam,&nbsp;Karthikeyan Elumalai","doi":"10.1002/ird3.70027","DOIUrl":"https://doi.org/10.1002/ird3.70027","url":null,"abstract":"<p>Artificial intelligence (AI) is changing how cancer is diagnosed, predicted, and treated, opening up new approaches to make cancer care more individualized. Rather than offering a broad but superficial overview, this review focuses on four cancers—lung, breast, brain (gliomas), and colorectal—for which AI was shown to be useful in the clinic. AI algorithms, specifically those using convolutional neural networks (CNNs), can enhance early diagnosis while realizing molecular profiling and treatment response assessment through quantitative imaging evaluations. Radiomics together with radiogenomics improves treatment accuracy through the assessment of imaging characteristics that help identify targeted genomic therapies. AI technologies can enhance tumor segmentation precision, stage determination, and target outlining capabilities, which enable adaptive radiation therapy. Initiatives that merge AI with images, clinical results, and genetic science information can deliver thorough personalized assessments that enhance treatment planning decisions. However, AI technology needs to overcome data quality issues, interpretability limitations, and generalizability challenges and needs to meet regulatory compliance requirements before achieving safe and fair implementation. The next phase of development will focus on federated learning to safeguard privacy while institutions collaborate, explainable AI to build transparent systems, and the fusion of diverse data types for comprehensive patient identification and real-time medical decision support through establishing digital twins for individualized treatment assessments. Precision oncology will be transformed by maturing innovations in predictive imaging that allow better timing of diagnosis while providing customized treatments to achieve improved medical results.</p>","PeriodicalId":73508,"journal":{"name":"iRadiology","volume":"3 4","pages":"261-278"},"PeriodicalIF":0.0,"publicationDate":"2025-07-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1002/ird3.70027","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144910074","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Undifferentiated Embryonal Sarcoma of the Liver","authors":"Lingqing Tang,&nbsp;Bin Yang","doi":"10.1002/ird3.70025","DOIUrl":"https://doi.org/10.1002/ird3.70025","url":null,"abstract":"<p>A 45-year-old male presented with upper abdominal pain that began 1 week ago, described as intermittent and dull. Physical examination revealed tenderness in the upper abdomen. The liver was palpable 10 cm below the right midclavicular line at the costal margin. Laboratory tests showed no significant abnormalities. The computed tomography image is shown in Figure 1a. The patient underwent right hemihepatectomy with caudate lobe resection. Histopathological findings are illustrated in Figure 1b. The diagnosis was a malignant tumor with necrosis, consistent with an undifferentiated sarcoma of the liver (UESL). During a 6-month follow-up, tumor metastasis was noted in the gastrointestinal space, along with multiple masses in the anterior left lobe of the liver and right renal space, indicating tumor recurrence. Dynamic axial contrast-enhanced CT scans showing mild heterogeneous enhancement of these lesions.</p><p>UESL, is an exceedingly rare malignant liver tumor. UESL ranks third in children, with adult occurrences being particularly uncommon.</p><p>Due to the rarity of UESL, imaging features lack specificity. CT scans revealed solid components often reside at the tumor margins with irregular septations and hemorrhage. Contrast-enhanced scans may demonstrate fast in fast out enhancement or delayed enhancement patterns. This case lacks typical imaging manifestations of enhancement, showing mild heterogeneous enhancement, which may be related to extensive hemorrhage and necrosis.</p><p><b>Lingqing Tang:</b> writing – original draft (lead), resources (equal). <b>Bin Yang:</b> resources (equal), writing – review and editing (lead).</p><p>The authors have nothing to report.</p><p>The patient has provided written informed consent prior to taking part in this study.</p><p>The authors declare no conflicts of interest.</p>","PeriodicalId":73508,"journal":{"name":"iRadiology","volume":"3 4","pages":"313-314"},"PeriodicalIF":0.0,"publicationDate":"2025-07-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1002/ird3.70025","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144909912","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Development of a Joint Prediction Model for Assessing the Severity of Hypertriglyceridemia-Induced Acute Pancreatitis","authors":"Junyao Long,&nbsp;Junjie Kuang,&nbsp;Zhuoya Ma,&nbsp;Zhuchun Guan,&nbsp;Qinghong Duan","doi":"10.1002/ird3.70024","DOIUrl":"https://doi.org/10.1002/ird3.70024","url":null,"abstract":"<div>\u0000 \u0000 \u0000 <section>\u0000 \u0000 <h3> Background</h3>\u0000 \u0000 <p>Patients with hypertriglyceridemia-induced acute pancreatitis (HTG-AP) have a high incidence of severe disease and a poor prognosis. This study aimed to construct a joint prediction model using multiple clinical and imaging indicators to assess the severity of HTG-AP.</p>\u0000 </section>\u0000 \u0000 <section>\u0000 \u0000 <h3> Methods</h3>\u0000 \u0000 <p>A retrospective analysis was conducted on 165 patients with HTG-AP, categorized into non-mild (<i>n</i> = 84) and mild (<i>n</i> = 81) groups. Clinical parameters were compared, and logistic regression was used to identify independent predictors. A joint prediction model was constructed and validated for stability and performance using receiver operating characteristic analysis, the bootstrap sampling method, the Hosmer–Lemeshow test, and the <i>Z</i>-test.</p>\u0000 </section>\u0000 \u0000 <section>\u0000 \u0000 <h3> Results</h3>\u0000 \u0000 <p>Significant intergroup differences were observed in lipid metabolism markers (total cholesterol [TC], high-density lipoprotein cholesterol [HDL-C], and low-density lipoprotein cholesterol [LDL-C]), pancreatic injury indicators (amylase [AMY] and lipase [LPS]), imaging characteristics (modified computed tomography severity index [MCTSI] score and liver computed tomography [CT] value), and hospitalization duration (<i>p</i> &lt; 0.05). The MCTSI score, liver CT value, TC level, and LDL-C level were identified as independent risk factors for non-mild HTG-AP. The joint model demonstrated superior performance (area under the curve [AUC] = 0.841) compared with individual predictors (<i>p</i> &lt; 0.05), with good calibration according to the Hosmer–Lemeshow test (<i>p</i> = 0.914) and stable performance validated by bootstrap sampling (ΔAUC = 0.001, <i>p</i> = 0.1531).</p>\u0000 </section>\u0000 \u0000 <section>\u0000 \u0000 <h3> Conclusion</h3>\u0000 \u0000 <p>The joint prediction model outperformed individual indicators such as the TC level, LDL-C level, MCTSI score, and liver CT value in assessing non-mild HTG-AP, offering enhanced clinical utility.</p>\u0000 </section>\u0000 </div>","PeriodicalId":73508,"journal":{"name":"iRadiology","volume":"3 4","pages":"302-310"},"PeriodicalIF":0.0,"publicationDate":"2025-07-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1002/ird3.70024","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144909988","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Bridging Clinical Knowledge and AI Interpretability in Thoracic Radiology","authors":"Mengze Xu","doi":"10.1002/ird3.70015","DOIUrl":"https://doi.org/10.1002/ird3.70015","url":null,"abstract":"&lt;p&gt;Yuan's study [&lt;span&gt;1&lt;/span&gt;] entitled “&lt;i&gt;Anatomic Boundary-Aware Explanation for Convolutional Neural Networks in Diagnostic Radiology&lt;/i&gt;” underscores a fundamental gap in existing XAI approaches: the neglect of clinical domain knowledge. Thoracic diseases primarily manifest within specific anatomical regions, such as the lung parenchyma. Yet, conventional XAI methods such as Grad-CAM or Integrated Gradients often highlight extraneous areas (e.g., medical devices, chest wall artifacts), leading to misinterpretations. By leveraging anatomic boundaries derived from a pretrained lung segmentation model, the authors enforce spatial constraints on CNN explanations, aligning them with clinically relevant regions. This innovation is particularly impactful for resource-limited settings, where annotations for fine-grained lesion localization are scarce.&lt;/p&gt;&lt;p&gt;The study's quantitative results are compelling: Across 72 scenarios involving 3 CNN architectures, 4 diseases, and 2 classification settings, the boundary-aware method outperformed baseline explanations in 71 cases. For example, in pneumothorax detection, the dice similarity coefficient (DSC) improved by up to 5.09% when integrating anatomic constraints. These findings validate the hypothesis that incorporating radiological expertise into XAI frameworks enhances explanation fidelity.&lt;/p&gt;&lt;p&gt;The paper's strengths lie in its plug-and-play design and transfer learning strategy. By decoupling lung segmentation from the CNN classifier, the authors avoid retraining on annotated target datasets, reducing computational and labeling costs. The use of publicly available segmentation datasets (e.g., Japanese Society of Radiological Technology) ensures reproducibility and scalability. However, this approach assumes minimal domain shift between external and target datasets. Future studies should evaluate robustness across diverse imaging protocols or patient populations, where anatomical variations (e.g., emphysematous lungs, postsurgical changes) might affect segmentation accuracy. Another notable aspect is the comprehensive evaluation of multiple XAI methods (saliency map, Grad-CAM, Integrated Gradients) and CNN architectures (VGG-11, ResNet-18, AlexNet) [&lt;span&gt;2&lt;/span&gt;]. The consistent improvements observed across these configurations suggest the boundary-aware framework is generalizable. However, the reliance on lightweight CNNs (e.g., VGG-11) raises questions about applicability to modern, deeper models (e.g., vision transformers), which may require different regularization strategies.&lt;/p&gt;&lt;p&gt;A limitation is the qualitative gap between improved metrics and clinical utility. Although intersection over union and DSC metrics quantify overlap with ground-truth lesions, they do not directly measure radiologists' trust in AI explanations. Future work should incorporate human-in-the-loop studies to assess how boundary-aware explanations influence diagnostic decisions and workflow efficiency.&lt;/p&gt;&lt;p&gt;Yuan's appro","PeriodicalId":73508,"journal":{"name":"iRadiology","volume":"3 4","pages":"311-312"},"PeriodicalIF":0.0,"publicationDate":"2025-06-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1002/ird3.70015","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144910359","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}